AMD Radeon 890M vs AMD Radeon Pro 450 Comparison

AMD Radeon Pro 450 vs AMD Radeon 890M

The AMD Radeon Pro 450 and the AMD Radeon 890M represent two distinct eras of mobile graphics, separated by nearly eight years of architectural evolution. The Pro 450 is a discrete MXM module from the GCN 4.0 generation, built for professional Mac workloads in 2016, while the 890M is a modern integrated graphics processor within the Strix Point mobile APU, built on RDNA 3.5. The database shows that the 890M delivers a dominant performance advantage in compute-oriented benchmarks, yet the Pro 450 holds its own in the context of its era, with a higher percentile ranking among all GPUs. This analysis breaks down where each part wins, the architectural differences that explain the gap, and the benchmark results that quantify the divide.

Where Each One Wins

The AMD Radeon 890M is the clear winner in raw compute throughput and modern API support. Across the two directly comparable benchmarks in the database, the 890M wins both, with no victories recorded for the Pro 450. The 890M’s advantage is most pronounced in Vulkan workloads, where it scores 40,808 against the Pro 450’s 10,518, a gap of roughly 74%. In OpenCL, the 890M scores 37,254 versus 9,002, again a massive margin. These results indicate that for any workload leveraging compute shaders, modern rendering pipelines, or Vulkan-based applications, the 890M is categorically superior.

The Pro 450, however, occupies a different niche. It is a discrete GPU with its own dedicated 2 GB of GDDR5 memory and a 128-bit memory bus, delivering 81.28 GB/s of bandwidth. This makes it self-sufficient in a way that an integrated part cannot match, since the 890M relies on system shared memory with bandwidth that is system dependent. The Pro 450’s 49th percentile among all GPUs is higher than the 890M’s 45th percentile, suggesting that despite its age, it sits in a slightly better position relative to its contemporaries when averaging across all recorded scores. The Pro 450 also posts a higher average benchmark score in the database at 10,804, compared to the 890M’s 9,210, though this average is skewed by the fact that the Pro 450 has only three recorded benchmarks, all of which are Geekbench compute tests, while the 890M has ten benchmarks across a wider range of test types, including several Passmark tests where integrated parts typically score lower.

For users running legacy DirectX 9 or DirectX 10 applications, the 890M’s Passmark scores offer some insight. The 890M scores 97 in Passmark DirectX 9, 33 in DirectX 10, 73 in DirectX 11, and 37 in DirectX 12. These are low absolute numbers, but they are the only data points available for older API performance. The Pro 450 has no equivalent direct comparisons in the database, so its legacy API performance cannot be assessed directly. The 890M is the winner for modern workloads, while the Pro 450’s edge is limited to its discrete memory architecture and slightly better percentile standing.

Architecture Differences

The architectural gap between these two GPUs is substantial. The Pro 450 is built on GlobalFoundries’ 14 nm process, while the 890M uses TSMC’s 4 nm node. That process difference alone explains much of the efficiency and density disparity. The Pro 450 packs 3,000 million transistors into a 123 mm² die, resulting in a transistor density of 24.4 million per square millimeter. The 890M integrates 34,000 million transistors into a 233 mm² die, yielding 145.9 million transistors per square millimeter, nearly six times denser. The 890M is an integrated part, meaning it shares the die with the CPU cores of the Strix Point APU, whereas the Pro 450 is a standalone MXM module.

The compute resource allocation is also heavily skewed. The Pro 450 has 640 shading units, 40 texture mapping units, and 16 raster output units. The 890M has 1,024 shading units, 64 TMUs, and 32 ROPs. The 890M also features 16 dedicated ray tracing cores, a feature entirely absent from the Pro 450. Pixel rate tells the story clearly: the 890M achieves 92.80 GPixel/s versus the Pro 450’s 12.80 GPixel/s. Texture rate follows the same pattern: 185.6 GTexel/s versus 32.00 GTexel/s. Floating point performance is where the gap becomes most extreme, with the 890M delivering 5.939 TFLOPS in FP32 and FP16, while the Pro 450 manages 1,024.0 GFLOPS in both. That is roughly 5.8 times more compute throughput for the 890M.

Clock speeds reflect the different design philosophies. The Pro 450 has no recorded base or boost clock in the database, but its memory runs at 1270 MHz with 5.1 Gbps effective. The 890M has a base clock of 400 MHz and a boost of 2900 MHz, a massive boost range that allows it to scale up under load. Memory architecture is fundamentally different: the Pro 450 uses dedicated 2 GB GDDR5 with a 128-bit bus, while the 890M uses system shared memory with no fixed size, type, or bus width, and bandwidth is system dependent. The Pro 450 supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3. The 890M supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, plus it carries the ray tracing hardware necessary for the Ultimate feature set. The 890M’s power envelope is notably lower at 15 W, compared to the Pro 450’s 35 W, and the 890M requires no power connectors while the Pro 450 has none listed either. The Pro 450 runs on PCIe 3.0 x8, while the 890M uses PCIe 4.0 x8. The 890M is an active product, released in July 2024, while the Pro 450 is end-of-life, released in October 2016.

Head-to-Head Benchmarks

The database contains two benchmarks where both GPUs have recorded scores, and the results are decisive. In Geekbench OpenCL, the 890M scores 37,254 against the Pro 450’s 9,002. The delta is 75.8% in favor of the 890M, meaning the integrated part is roughly four times faster in this compute test. This is a massive margin, one that reflects not just the additional shading units and higher clocks, but also the architectural efficiency of RDNA 3.5 over GCN 4.0. OpenCL is a general-purpose compute API, and the 890M’s near 6 TFLOPS of FP32 throughput simply overwhelms the Pro 450’s 1.0 TFLOPS.

In Geekbench Vulkan, the 890M scores 40,808 while the Pro 450 scores 10,518. The delta is 74.2% in favor of the 890M. Vulkan is a low-level graphics and compute API, and the 890M’s support for Vulkan 1.4, combined with its modern architecture, gives it a clear advantage. The Pro 450 supports Vulkan 1.3, which is not a limitation in this test, but the raw hardware difference is too large to overcome. The 890M’s 16 ray tracing cores are not directly benchmarked in these tests, but they contribute to the architectural superiority that shows up in the compute scores.

For context, the Pro 450’s nearest rivals in the database are the NVIDIA Quadro K2200, which scores 10,761 and is 0.4% ahead, and the NVIDIA GeForce MX350, which scores 10,883 and is 0.7% ahead. The Pro 450 sits between these two, slightly behind the MX350 and slightly ahead of the Quadro K2200. The 890M’s nearest rivals are the AMD Radeon Vega 8, which scores 9,221 and is 0.1% ahead of the 890M, and the NVIDIA GeForce GTX 960, which scores 9,273 and is 0.7% ahead. This means that despite the 890M’s massive compute lead over the Pro 450, its overall average benchmark score is actually lower, because the Passmark tests in its suite drag down the average. The Pro 450’s average of 10,804 is higher than the 890M’s 9,210, but this is a function of test selection rather than real-world performance parity.

The takeaway is that in any compute-heavy or modern graphics workload, the 890M wins by a landslide. The Pro 450’s only formal advantage in the recorded data is its higher percentile rank and higher average score, both of which are artifacts of a smaller, more focused benchmark set. When the two are directly compared on the same tests, there is no contest.

FAQ

Q: Which GPU has more shading units?

A: The AMD Radeon 890M has 1,024 shading units, while the AMD Radeon Pro 450 has 640 shading units.

Q: How much faster is the 890M in Geekbench OpenCL?

A: The 890M scores 37,254 in Geekbench OpenCL, compared to the Pro 450’s 9,002, a difference of 75.8% in favor of the 890M.

Q: Does the Radeon Pro 450 support ray tracing?

A: No. The Pro 450 has no ray tracing cores listed in the database. The Radeon 890M has 16 ray tracing cores.

Q: What is the memory configuration of the Radeon Pro 450?

A: The Pro 450 has 2 GB of GDDR5 memory on a 128-bit bus, with 81.28 GB/s of bandwidth. The 890M uses system shared memory with no fixed size or bandwidth.

Q: Which GPU has a higher percentile ranking among all GPUs?

A: The Pro 450 has a 49th percentile ranking, while the 890M has a 45th percentile ranking.

Q: What are the floating point performance figures for each GPU?

A: The Pro 450 delivers 1,024.0 GFLOPS in both FP32 and FP16. The 890M delivers 5.939 TFLOPS in both FP32 and FP16.

The Verdict

The data leaves no room for ambiguity in raw performance: the AMD Radeon 890M is the superior part for any modern workload. It wins both head-to-head benchmarks by margins of roughly 74% to 76%, it has more shading units, more TMUs, more ROPs, dedicated ray tracing cores, and nearly six times the FP32 throughput. It does all of this at a lower power draw of 15 W, compared to the Pro 450’s 35 W, and it supports newer API versions including DirectX 12 Ultimate and Vulkan 1.4. For anyone choosing between these two for a current system, the 890M is the obvious pick.

The Pro 450 is not without its merits, but those merits are contextual. It has a higher percentile ranking, 49th versus 45th, and a higher average benchmark score, 10,804 versus 9,210, but both figures are influenced by the limited set of benchmarks recorded for the Pro 450. Its dedicated 2 GB of GDDR5 memory with 81.28 GB/s of bandwidth gives it a self-contained memory subsystem, which can be an advantage in systems with slow or insufficient system memory. However, that advantage is erased by the 890M’s far superior compute resources and architectural efficiency.

Given that the Pro 450 is end-of-life and the 890M is an active product with a 2024 release date, the practical recommendation is straightforward. The 890M is the right choice for modern gaming, compute, and any application that can leverage Vulkan or DirectX 12 Ultimate features. The Pro 450 should only be considered for legacy professional environments where its specific MXM form factor or discrete memory architecture is required, and where the software in question is optimized for GCN 4.0 hardware. For everyone else, the benchmark results are unambiguous: the 890M is the faster, more capable, and more efficient GPU.